A resource-constrained stochastic scheduling algorithm for homeless street outreach and gleaning edible food

Abstract

We developed a common algorithmic solution addressing the problem of resource-constrained outreach encountered by social change organizations with different missions and operations: Breaking Ground -- an organization that helps individuals experiencing homelessness in New York transition to permanent housing and Leket -- the national food bank of Israel that rescues food from farms and elsewhere to feed the hungry. Specifically, we developed an estimation and optimization approach for partially-observed episodic restless bandits under $k$-step transitions. The results show that our Thompson sampling with Markov chain recovery (via Stein variational gradient descent) algorithm significantly outperforms baselines for the problems of both organizations. We carried out this work in a prospective manner with the express goal of devising a flexible-enough but also useful-enough solution that can help overcome a lack of sustainable impact in data science for social good.

Figures (8)

• Figure 1: Restless Bandit Stein Sampling (RB-Stein)
• Figure 2: TSDE version of RB-Stein. $N^{i}_{t}(s,a)$ is the number of times arm $i$ has visited $(s,a)$ up to time $t$. $\Tilde{\mathcal{S}}^{M}_{t}$ and $\tilde{r}^{M}_{t}$ are the state and reward vectors of the top $M$ arms observed at time $t$.
• Figure 3: Transition Estimation Algorithm Description. $df(\vert S \vert) := \vert \mathcal{S} \vert \times (\vert \mathcal{S} \vert -1)$.
• Figure 4: RB-Stein with Mirror SVGD and Dynamic Episodes (TS MCR) : Leket. From left to right: $N = 10$, $N=50$, and $N=100$ arms.
• Figure 5: RB-Stein with Mirror SVGD and Dynamic Episdoes (TS MCR): Breaking Ground. From left to right: $N = 10$, $N=50$, and $N=100$ arms.